@inproceedings{chen-etal-2026-ticd,
title = "{TICD}: Time-Aware Internal Contrastive Decoding for Enhancing Factuality in Diffusion Language Models",
author = "Chen, Zike and
Dou, Hongkun and
You, Mukang and
Deng, Yue and
Li, Jiajun",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.1303/",
pages = "27558--27574",
ISBN = "979-8-89176-462-0",
abstract = "Diffusion large language models (dLLMs) generate text through iterative denoising, but they remain prone to factual errors in knowledge-intensive tasks. Through layer-wise and timestep-wise probing, we observe a spatiotemporal stratification pattern in dLLMs: structural-token predictions tend to converge earlier across layers, whereas factual-token predictions rely more heavily on deep layers and become reliable only after sufficient denoising progress. Motivated by this observation, we propose Time-Aware Internal Contrastive Decoding (TICD), a training-free decoding method that calibrates final-layer predictions using entropy-gated intermediate-layer references only during semantically reliable denoising stages. This design uses reliable final-intermediate prediction gaps as correction signals to strengthen factual predictions at the final layer. Experiments on LLaDA-8B and Dream-7B across factuality and reasoning benchmarks show consistent improvements over vanilla decoding."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="chen-etal-2026-ticd">
<titleInfo>
<title>TICD: Time-Aware Internal Contrastive Decoding for Enhancing Factuality in Diffusion Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zike</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hongkun</namePart>
<namePart type="family">Dou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mukang</namePart>
<namePart type="family">You</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yue</namePart>
<namePart type="family">Deng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiajun</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Diffusion large language models (dLLMs) generate text through iterative denoising, but they remain prone to factual errors in knowledge-intensive tasks. Through layer-wise and timestep-wise probing, we observe a spatiotemporal stratification pattern in dLLMs: structural-token predictions tend to converge earlier across layers, whereas factual-token predictions rely more heavily on deep layers and become reliable only after sufficient denoising progress. Motivated by this observation, we propose Time-Aware Internal Contrastive Decoding (TICD), a training-free decoding method that calibrates final-layer predictions using entropy-gated intermediate-layer references only during semantically reliable denoising stages. This design uses reliable final-intermediate prediction gaps as correction signals to strengthen factual predictions at the final layer. Experiments on LLaDA-8B and Dream-7B across factuality and reasoning benchmarks show consistent improvements over vanilla decoding.</abstract>
<identifier type="citekey">chen-etal-2026-ticd</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1303/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>27558</start>
<end>27574</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T TICD: Time-Aware Internal Contrastive Decoding for Enhancing Factuality in Diffusion Language Models
%A Chen, Zike
%A Dou, Hongkun
%A You, Mukang
%A Deng, Yue
%A Li, Jiajun
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F chen-etal-2026-ticd
%X Diffusion large language models (dLLMs) generate text through iterative denoising, but they remain prone to factual errors in knowledge-intensive tasks. Through layer-wise and timestep-wise probing, we observe a spatiotemporal stratification pattern in dLLMs: structural-token predictions tend to converge earlier across layers, whereas factual-token predictions rely more heavily on deep layers and become reliable only after sufficient denoising progress. Motivated by this observation, we propose Time-Aware Internal Contrastive Decoding (TICD), a training-free decoding method that calibrates final-layer predictions using entropy-gated intermediate-layer references only during semantically reliable denoising stages. This design uses reliable final-intermediate prediction gaps as correction signals to strengthen factual predictions at the final layer. Experiments on LLaDA-8B and Dream-7B across factuality and reasoning benchmarks show consistent improvements over vanilla decoding.
%U https://aclanthology.org/2026.findings-emnlp.1303/
%P 27558-27574
Markdown (Informal)
[TICD: Time-Aware Internal Contrastive Decoding for Enhancing Factuality in Diffusion Language Models](https://aclanthology.org/2026.findings-emnlp.1303/) (Chen et al., Findings 2026)
ACL